Intel Xe GPU Flagship To Be Made On TSMC’s 6nm And 3nm Fabrication Node?

Intel has long struggled to evolve from the 14nm Fabrication Node to the 10nm Production Process. But now it seems the company might abandon the 10nm process too. As we had reported earlier, Intel is believed to be considering directly jumping to the sub 7nm Fabrication Process. Interestingly, the company might not partner with Samsung. Instead, Intel is rumored to be considering Taiwan’s TSMC for its 6nm and 3nm processes for high-performance editions of its Xe Graphics Chips.

Intel is believed to be considering dropping the 10nm fabrication node, but the decision could be restricted to its upcoming Xe Graphics Chips. While the current generation of Xe DG1 GPU could be mass-produced on the 10nm Fabrication Node, the next generation of the Xe Graphics would be manufactured on the sub 7nm Production Process. Interestingly, Intel could approach TSMC as the latter has fast-tracked the development of the new production processes which significantly reduce the die size. It is important to note that neither Intel nor TSMC has confirmed or denied the reports yet. Hence the news might be just a rumor. Nonetheless, given the long struggle Intel has faced moving beyond the archaic 14nm Fabrication Node, the company might just outsource the production of Xe DG2 GPU.

Intel Approaching TSMC To Produce Its Next-Gen Xe GPUs On The Sub 7nm Fabrication Process?

Intel is yet to commercially launch its Xe DG1 GPU. In fact, the company has only mentioned the existence of Xe GPUs. Intel is expected to conduct commercial trials of the Xe DG1 this year. The Graphics Card based on the Xe DG1 Architecture is based on the newly finalized 10nm process.

The Intel Xe DG1 GPU features just 96 EUs. It consumes merely 25W of power. In other words, the performance and power consumption of the Intel GPU is certainly higher than the currently popular NVIDIA MX 250 Series. However, experts claim the 10nm Intel Xe DG1 GPU is essentially the TGL iGPU in a discrete form factor. It is quite apparent that Intel is merely experimenting with the Xe DG1 GPU before manufacturing and commercially launching the high-performance edition of the Xe GPU.

The rumor about abandoning the 10nm Fabrication Node in favor of TSMC’s 7nm EUV process and then continuing with the latter’s even miniaturized production process makes perfect sense. Intel’s CFO has openly complained that the production yields from the 10nm process being suboptimal. Consequentially, the platform will be quite lower in profitability than the multi-generation old 22nm process. Simply out, in its current iteration, Intel’s own 10nm Fabrication Process might not be able to ever offer mass-produced Xe GPUs. Hence, the company will be obviously looking for better production techniques with much better yields, and in turn, profits.

Intel’s Technology and Manufacturing Group (TMG) has long struggled with a commercially viable production process for the sub 14nm production process. It is no wonder the company is still trying to sell CPUs manufactured on the 14nm Production Process, while AMD, its primary competitor, has moved all its product lineup to the 7nm production process. Incidentally, AMD relies on TSMC for its CPUs and GPUs.

According to the rumors, Intel will hand over the production of Xe DG2 Graphics Chips to TSMC. The Taiwanese giant will develop and manufacture the next generation of high-performance editions of Xe Graphics on sub 7nm, which 6nm and even 3nm production processes. It is important to note that Intel has confirmed that Ponte Vecchio will be manufactured on the 7nm process. Hence it is quite likely that Intel will rely on two manufacturing plants for its Xe GPUs in the near future.


Alap Naik Desai

A B.Tech Plastics (UDCT) and a Windows enthusiast. Optimizing the OS, exploring software, searching and deploying solutions to strange and weird issues is Alap's main interest.